Dexcom system design query

Hi All

I've been researching the Dexcom system design. As far as I can tell, the transmitter is a transmitter ONLY and the receiver is a receiver ONLY.

What this means is that when you issue a command to the receiver, it doesn't actually go anywhere near the transmitter/sensor, but is processed entirely (faked up) in the receiver.

This covers things like resetting the sensor, turning off the sensor, calibrations etc.

Or is this wrong?

The receiver is labelled receive only on the back. But the chips used are capable of being receivers and transmitters.

Anyone know for sure? Surely the FCC would need to know for approvals.

What have you expected? The transmitter is just sending its data every 5 minutes. It is the receiver that will use its algorithms to calculate glucose numbers based on this raw data, the calibrations and interpolation. The intelligence is in the interpretation of the data so there is no need to control the sensor or transmitter directly.

Hi Holger. I'm just used to most comms systems being 2 way. There are several features in the remote that are labelled in such a way as to give the impression that the receiver sends commands to the sensor. These are items like "Stop sensor", etc. Really I guess this just starts and stops the processing in the receiver.

This is all fine. The issues we see in using the device are things like:

1. we have to follow our 3 yo around all day with the receiver (or strap it to him and risk him breaking it) to prevent from dropping data points. There are several ways dexcom could ameliorate this, even with a unidirectional system - e.g. always just send the last 10 readings.

2. resetting the device makes it off line for 2 hours, when really if the receiver could just be convinced to keep going with existing calibration data, then there could be no down time.

3. Out of range. This just means some timer expired in the receiver, since there's no carrier to detect (except during the very short window). So it is confusing when it takes a long time to show back in range, even when the receiver is close to the transmitter. It makes people think something else is wrong (which may cause them to do bad things, like reset the receiver, or change the transmitter or something).

I understand FDA approval requirements have caused a lot of operational compromises, but for those of us operating above the lowest common denominator, we miss out on quite a few things.

Even being able to scroll back more than a day (I can imagine some person with low intelligence having issues because they don't notice they are scrolled back, so nobody gets scrolling). Or zoom in. We try to keep in range 4 - 7.7, yet it always shows the range up to 25 or so, most of the screen is therefore not used, and you lose precision. This could be resolved and remain bulletproof for FDA approval with some automatic setting of the Y axis range.

I'd also love to see a dynamic schedule for tests. When stable, 1/5min is too often, and when moving quickly it's too slow. There should be an option to increase sample rate when the values are changing (e.g. automatically). I think this overall would not result in reduced operational life through enzyme use as the fast times would balance out the slow ones.

I'd also love to be able to know when to expect the next result, and see timestamps on results (e.g. a cursor).

We also desperately need a remote monitoring option, for when my son is at kindergarten. I'm currently looking into repeater options, since the TI chipset etc used is readily available at very low cost.

Hazard of being an engineer is wanting to fix things...

also I think when you put in a new sensor, that 2 hours spent "initialising it" is actually just 2 hours locking out the results and doing nothing (except showing an hourglass etc). From what I understand, the transmitter is sending results prior to that, but they are being ignored. Probably trials found initial results unreliable so they lock them out, and chose 2hrs as a compromise.

The encymatic reaction in the sensor can not be stopped. It will create data all the time. This data is transmitted every 5 minutes. Of course it would have been nice to buffer the data in the transmitter but for that you need two way communication (OK, repeat from the receiver). That would make the transmitter more complex and thirsty for power. For your 3 yo it would be ideal no doubt. For their design they focused on people carrying the receiver with them. I am not even sure if the Dexcom is officially for children. This would not hinder me and I just argue here that Dexcom has focused on a different group of users.

We have this constant stream of data and the device is still showing "???" and is just collecting the data in the background. It is kind of absurd but for me it shows that the data is not very reliable. They do not generate good data every 5 minutes. Thus they interpolate to eleminate all the random noise they have collected with the data. I think they sticked to the long "initialization" phase to be legally and technically on the safe side. Very likely they have choosen a timespan that results in the best correlation between blood glucose and sensor result. Usually this span is the outcome of 100 or more people wearing the Dexcom in lab conditions (the trials you have mentioned). On top of that there are strict FDA requirements and tight regulations for medical devices but maybe I am too forgiving here.

the chipset used in the transmitter already has the capability to receive, so the existing hardware could do it, it even has the memory that could be used for the buffer.

If it did a simple ping prior to sending, and waited for a response from the receiver, it could know if the receiver was in range and either send or buffer. Or it could just always send the last 10 or some number of readings, then you'd get the data next time you came in range.

Anyway, been doing a fair bit of research reading dexcom patents. Very interesting sensor design, I don't feel so bad about the cost of the sensor now, knowing it has platinum filament, silver electrode, semi-permeable membranes, and enzyme and oxidation layers. Some of their newer patents pending are very interesting, pointing the way to their G5 and what looks like it might be their G6 product (at least it's implantable).